Allow using comparison operator symbols as tests

[davidism]

I added the Python short names for the operators as well as the symbols. I kept the long names for the three existing tests.

parse_test is adjusted to allow comparison symbols as test names, so that they can be used in direct tests as well as filters.

{{ x is <= 5 }} isn't very useful, but it's allowed for consistency
{{ x is le 5 }}
{{ x|select('<=', 5) }}
{{ x|select('le', 5) }}

Todo:

• unit tests
• documentation
• changelog

closes #664

[ThiefMaster]

Is it really a good idea to allow symbol operators in "normal" tests, i.e. for stuff like foo is == 'bar'? That looks completely awful to me to be honest. I'd rather allow an "extended set of tests" for the places where you use them as strings, i.e. select, reject and their ...attr variants.

This would also add the advantage that editors and IDEs won't have to deal with new syntax when highlighting/checking Jinja code - I'm quite quire that PyCharm is not the only IDE that currently shows {% if foo is == 'bar' %} as a syntax error.

[davidism]

@ThiefMaster I thought it was a bit weird too, but I figured people would get confused by the inconsistency between is and select otherwise. The docs already call out equalto as being useless in is.

Since there was no expectation before that symbols could be used, we could just document them as a special case in the select docs, rather than in the general test docs.

Or we could still allow them in both locations, but only document them in select. I can open a PyCharm issue for the syntax, although we both know it could take a while to get in.

[davidism]

Going to take out the is <= support, since I agree it looks weird. Here's the patch if we decide we need this later.

diff --git a/jinja2/parser.py b/jinja2/parser.py
index 6d1fff6a..8741520a 100644
--- a/jinja2/parser.py
+++ b/jinja2/parser.py
@@ -825,10 +825,19 @@ def parse_test(self, node):
             negated = True
         else:
             negated = False
-        name = self.stream.expect('name').value
-        while self.stream.current.type == 'dot':
-            next(self.stream)
-            name += '.' + self.stream.expect('name').value
+
+        # If the next token is a comparison operator, treat it as a name.
+        # This allows using the operator symbols in the select filter without
+        # breaking unexpectedly when used in a test node.
+        if self.stream.current.type in _compare_operators:
+            name = next(self.stream).value
+        else:
+            name = self.stream.expect('name').value
+
+            while self.stream.current.type == 'dot':
+                next(self.stream)
+                name += '.' + self.stream.expect('name').value
+
         dyn_args = dyn_kwargs = None
         kwargs = []
         if self.stream.current.type == 'lparen':